How to Decide Whether to Front a Rebate for a Customer

Why this matters

Sooner or later a customer asks you to take the rebate off the price now instead of making them wait for it. It sounds like a service and it feels like a closing tool. It is neither. It is an unsecured advance you are making against a decision a stranger will make weeks from now, repaid by a party you have no contract with, at no interest, with no security. Shops that do it well underwrite it like a lender. Shops that do it badly discover the exposure only after a run of denials, by which point the money is already out the door across a dozen jobs. Underwrite it before the first one, not after the twelfth.

Call it what it is

When you front a rebate you take on three things at once, and it helps to name them separately because they fail separately.

• Credit risk. The claim may be denied, in which case you have discounted the job by the incentive amount and received nothing for it.
• Float. Even on a claim that pays perfectly, your cash is out from the day of install until the day the administrator pays, which on most programs is measured in months.
• Collection risk, but only in one structure. If the program pays the customer rather than you, fronting means chasing your money through a customer who has already been paid and already considers the job closed. That is the worst version of this and it is also the most common one shops stumble into.

Notice that customer creditworthiness barely appears on that list. You are not underwriting the customer, you are underwriting a program's approval decision. The exception is exactly the structure named in the third bullet, and the fact that it turns the deal into a customer-credit decision is the reason to refuse it rather than a reason to check their credit.

Three arrangements, compared

The third row is the one to strike from your options entirely. There is no version of it that is not a personal loan dressed as a discount.

The four gates, all of which must pass

These are conjunctive. All four, on every job, before any money is fronted.

Gate 1: the program permits you to be the payee. Read the participation agreement for an assignment-of-benefit provision or a contractor-of-record model, and use the program's own assignment form if one exists. A private assignment your office wrote is not binding on an administrator who never agreed to it. If the program pays the customer and offers no assignment path, stop here. This gate alone eliminates most requests.

Gate 2: the approval is either in hand or measurable. A written pre-approval, unexpired, is the strongest possible position and effectively removes the credit risk. Absent that, you need your own claim log on that specific program with at least 25 closed claims before you can treat your paid rate as a real number. Below that count you are not measuring an approval rate, you are looking at noise, and the honest answer is do not front.

Gate 3: you know the clawback terms and can survive them. Many programs reserve the right to audit after payment and recover money where the installation fails verification. If you have been paid by assignment, the recovery comes back to you, not to the customer, and it can arrive long after the job closed. Know the audit window and keep the file until it expires.

Gate 4: you have the cash for it, at the book level, not the job level. Covered below, and it is the gate that fails most often for shops that pass the other three.

What fronting actually costs, in two parts

Both parts have to be priced, and they are different costs.

Part one, the expected loss. Take your own paid rate on that program from your own log. Say a shop's log shows 52 closed claims, 48 paid and 4 permanently denied. That is about 92 percent paid (48 of 52) and about 8 percent denied (4 of 52). If the incentive on this program averages about a fifth of the ticket, the expected loss on each fronted job is 8 percent of a fifth of the ticket, which is 1.6 percent of the ticket.

That sounds trivial until you set it against margin. If the shop's net margin on this work is 8 percent of the ticket, then 1.6 points of expected loss is one fifth of the net margin, given away on every job it fronts, before the float is counted at all. A 92 percent approval rate feels excellent. Losing a fifth of net margin does not. Both statements describe the same book.

Part two, the float. From the same log, median time from submission to payment was 63 days, roughly 9 weeks, with the longest single claim at 147 days. If the shop fronts 2 jobs a week and each fronted amount sits out for a median of 9 weeks, the steady-state balance is about 18 jobs' worth of incentive outstanding at any moment (2 times 9). At a fifth of a ticket each, that is about 3.6 tickets' worth of cash permanently parked outside the business, not counting the tail cases running out past 20 weeks.

Nothing about that 3.6 is a loss. It is cash that exists and that you cannot use, which is a different problem with the same symptoms. See the sibling card on cash versus profit if that distinction is not already reflexive.

The portfolio limit, and why it binds before the customer does

Set a cap on total open fronted exposure and enforce it as a hard stop.

A workable default: cap open fronted exposure at the equivalent of 4 tickets, measured across all open fronted claims at any moment, and raise the cap by 1 ticket for every additional 25 closed claims of clean history on that program. The unit of analysis is the whole open book at a point in time, not the individual job, and the step size is 1 ticket per 25 claims so the cap grows on evidence rather than on optimism. Lower it by the same step after any season where the paid rate on that program falls below your prior measured rate.

Adopt the numbers as written and tune them to your own cash position; what matters more than the exact cap is that a cap exists, is measured at the book level, and is checked before each individual decision rather than after.

This is the gate that surprises owners, because it declines customers who are perfectly fine. The constraint that binds is your book, not the borrower. A shop at its cap declines the next request not because that customer is risky but because it has already committed the capacity it set aside for this. That is what a limit is for, and it is the reason to set it while nothing is wrong.

Worked example: two customers, one book

Same shop, same week, same job type, using the log above: 92 percent paid, median 9 weeks, incentive about a fifth of the ticket, cap of 4 tickets, current open fronted exposure of 3.6 tickets.

Customer A. The program has a contractor-of-record model with a published assignment form, and the shop holds a written pre-approval for this job with an expiry 6 weeks out. Gates 1, 2 and 3 pass cleanly: the shop can be the payee, the approval is in hand rather than predicted, and the audit window is known and short enough to track.

Gate 4 is where it lands. Open exposure is 3.6 tickets against a 4-ticket cap, so this job's fifth of a ticket takes the book to 3.8 tickets, still under the cap. The shop fronts it, files the assignment on the program's own form, schedules the install well inside the pre-approval's 6-week expiry, and submits within its 5-business-day internal window.

Note what carried the decision. It was not that Customer A seemed reliable. It was the pre-approval and the assignment path. Had gate 4 failed instead, with the book already at 4 tickets, the correct answer would have been to decline this identical, well-documented job. That is not a defect in the rule, it is the rule working.

Customer B. Same job, same week, but the program pays the customer directly and has no assignment provision. The customer is a long-standing account with a perfect payment history and asks pointedly why the shop can do it for others and not for them.

Gate 1 fails, so the analysis ends. Every other factor is favorable and none of them matter, because the structure would leave the shop collecting an amount from a customer who has already banked the program's check and mentally closed the job. The customer's excellent history is the strongest argument against making an exception, since exceptions get made for good customers and that is precisely how a shop ends up with a portfolio of informal personal loans.

The answer to the pointed question is structural and easy to say plainly: on this program the utility pays you, not us, so there is nothing for us to be assigned. That is a fact about the program, not a judgment about the customer, and customers accept it because it is true.

Do not accidentally become a consumer lender

If you structure fronting as the customer owing you the incentive amount later, rather than as an assignment of the program's payment, you have created a deferred payment obligation. Consumer credit arrangements come with federal disclosure duties under the Truth in Lending Act as implemented by Regulation Z, which generally reach credit extended to a consumer for personal, family or household purposes that is either subject to a finance charge or payable by written agreement in more than four installments. A shop improvising a payment arrangement at a kitchen table is not thinking about coverage tests, which is exactly the problem.

Keep it clean: use assignment where the program provides it, decline where it does not, and route any genuine financing need to a third-party lender who is set up for it.

What to offer instead when you decline

Declining to front should not end the conversation empty-handed. There are three substitutes and all three are better than an informal loan.

• Third-party financing. A lender whose business is consumer credit, presented as a monthly commitment the customer evaluates on its own terms, entirely separate from whether an incentive ever pays.
• A deposit and progress structure. Restructuring when your money arrives is not the same as reducing what you are owed, and it solves a genuine cash-timing problem for the customer without any exposure to a program's decision.
• Doing the paperwork well and fast. The most valuable thing you can hand a customer who cannot wait is a claim that files within 5 business days of commissioning with no defects, which is often several weeks faster than the same claim would have moved had they filed it themselves.

References

• Truth in Lending Act, 15 U.S.C. 1601 and following, as implemented by Regulation Z, 12 CFR Part 1026, on coverage of consumer credit arrangements
• Program participation agreement and assignment-of-benefit provisions, which determine whether a shop can lawfully be the payee of record
• See related: Cash vs Profit: Why They're Different; The Cash Flow Shape of Rebate-Heavy Work; The Promise a Shop Cannot Make About Someone Else's Money